Re: [PATCH v2 10/10] usb: chipidea: tegra: Add USB_TEGRA_PHY module to driver's dependencies

[Dmitry Osipenko]

23.12.2019 09:40, Peter Chen ÐÐÑÐÑ:
> On 19-12-20 07:31:08, Dmitry Osipenko wrote:
>> 20.12.2019 06:56, Peter Chen ÐÐÑÐÑ:
>>> On 19-12-20 04:52:38, Dmitry Osipenko wrote:
>>>> Now, when ci_hdrc_tegra kernel module is loaded, the phy_tegra_usb module
>>>> is loaded too regardless of kernel's configuration. Previously this
>>>> problem was masked because Tegra's EHCI driver is usually enabled in
>>>> kernel's config and thus PHY driver was getting loaded because of it, but
>>>> now I was making some more thorough testing and noticed that PHY's module
>>>> isn't getting auto-loaded without the host driver.
>>>>
>>>> Note that ChipIdea's driver doesn't use any of the exported functions of
>>>> phy_tegra_usb module and thus the module needs to be requested explicitly.
>>>>
>>>> Signed-off-by: Dmitry Osipenko <digetx@xxxxxxxxx>
>>>> ---
>>>>  drivers/usb/chipidea/Kconfig         | 1 +
>>>>  drivers/usb/chipidea/ci_hdrc_tegra.c | 6 ++++++
>>>>  2 files changed, 7 insertions(+)
>>>>
>>>> diff --git a/drivers/usb/chipidea/Kconfig b/drivers/usb/chipidea/Kconfig
>>>> index ae850b3fddf2..d53db520e209 100644
>>>> --- a/drivers/usb/chipidea/Kconfig
>>>> +++ b/drivers/usb/chipidea/Kconfig
>>>> @@ -7,6 +7,7 @@ config USB_CHIPIDEA
>>>>  	select RESET_CONTROLLER
>>>>  	select USB_ULPI_BUS
>>>>  	select USB_ROLE_SWITCH
>>>> +	select USB_TEGRA_PHY if ARCH_TEGRA
>>>>  	help
>>>>  	  Say Y here if your system has a dual role high speed USB
>>>>  	  controller based on ChipIdea silicon IP. It supports:
>>>> diff --git a/drivers/usb/chipidea/ci_hdrc_tegra.c b/drivers/usb/chipidea/ci_hdrc_tegra.c
>>>> index 7455df0ede49..8bc11100245d 100644
>>>> --- a/drivers/usb/chipidea/ci_hdrc_tegra.c
>>>> +++ b/drivers/usb/chipidea/ci_hdrc_tegra.c
>>>> @@ -53,6 +53,12 @@ static int tegra_udc_probe(struct platform_device *pdev)
>>>>  	struct tegra_udc *udc;
>>>>  	int err;
>>>>  
>>>> +	if (IS_MODULE(CONFIG_USB_TEGRA_PHY)) {
>>>> +		err = request_module("phy_tegra_usb");
>>>> +		if (err)
>>>> +			return err;
>>>> +	}
>>>> +
>>>
>>> Why you do this dependency, if this controller driver can't
>>> get USB PHY, it should return error. What's the return value
>>> after calling below:
>>>
>>> 	udc->phy = devm_usb_get_phy_by_phandle(&pdev->dev, "nvidia,phy", 0);
>>
>> It returns -EPROBE_DEFER when phy_tegra_usb isn't loaded.
>>
>> So if you'll do:
>>
>> # rmmod ci_hdrc_tegra; rmmod ci_hdrc; rmmod phy_tegra_usb;
>> # modprobe ci_hdrc_tegra
>> # lsmod
>> Module                  Size  Used by
>> ci_hdrc_tegra          16384  0
>> ci_hdrc                45056  1 ci_hdrc_tegra
>>
>> After this patch:
>>
>> # rmmod ci_hdrc_tegra; rmmod ci_hdrc; rmmod phy_tegra_usb;
>> # modprobe ci_hdrc_tegra
>> # lsmod
>> Module                  Size  Used by
>> Module                  Size  Used by
>> phy_tegra_usb          20480  1
>> ci_hdrc_tegra          16384  0
>> ci_hdrc                45056  1 ci_hdrc_tegra
> 
> I wonder why the driver needs such dependency? If there are two phy
> drivers could work with this controller driver, you may request two
> modules.

Well, if somebody wants to use some PHY driver other than the upstream's
standard one, then that person could simply load the custom driver
module first, such that it will bind to the PHY's device first.

It is also possible to manually unbind the standard driver from PHY's
device and then bind whatever driver you want.

> Doesn't such dependency should be done by the board
> level script?

This patch only improves the default behaviour that is common for all
NVIDIA Tegra boards, it doesn't prevent from doing any special
customizations.

Perhaps the Kconfig change could be dropped from this patch in order to
provide a bit more flexibility in regards to kernel's configuration, but
I'm very doubtful that realistically anyone would want to replace the
default driver with anything else on Tegra. The Kconfig change also puts
ChipIdea's UDC driver in line with the Tegra's EHCI driver that selects
USB_TEGRA_PHY, please see drivers/usb/host/Kconfig.

> Do you know are there any other drivers do such things?

I don't think that any of the USB host drivers are currently doing such
things, but in general there are quite a lot of drivers in kernel that
are using request_module [1].

[1] https://elixir.bootlin.com/linux/latest/ident/request_module

Please note that drivers/usb/host/ehci-tegra.c uses exported symbols
from usb/phy/phy-tegra-usb.c and that is why the EHCI driver doesn't
need to explicitly load the phy_tegra_usb module, the load happens
automatically because of the missing symbols.

Also, please note that it is possible to squash the Tegra EHCI driver
into ci_hdrc_tegra.c and then the explicit dependency on the
phy_tegra_usb won't be needed anymore since it will be replaced with an
implicit dependency. We (me and Peter Geis) already had some
experimental patches that do the successful squashing of the drivers,
but looks like Peter got sidetracked for a more important things for
now, we'll probably return to that work later on.